Jlf^M.

H A V

LIBRARY!UNIVERSITY OFCALIFORNIASANTA CRUZ

Digitized by tine Internet Arciiive

in 2007 witii funding from

IVIicrosoft Corporation

littp://www.arcliive.org/details/fossilturtlesofnOOIiayoricli

THE

FOSSIL TURTLESOF

NORTH AMERICA

BY

OLIVER PERRY HAY

WASHINGTON, D. C.Published by the Carnegie Institution of Washington

1908

Carnegie Institution of Washington, Publication No. 75

ISAAC H. BLANCHARD COMPANY

NEW YORK

QEg(oZ

H3PREFACE.

In this treatise on the Fossil Turtles of North America there are described

266 species, of which 76 are regarded as hitherto unknown to science. In thepreparation of this work the writer has had access to most of the collections whichcontain remains of North American fossil turtles. The most important of thesecollections is that of the American Museum of Natural History, in New York.In this are found many of the specimens described by Professor E. D. Cope,including many of his types. In addition to these, large numbers of turtles havebeen brought together by the expeditions sent out by this museum during the pastfifteen years. Free access has been given the writer to the materials in the United

States National Museum, where there are many of the specimens described byDr. Joseph Leidy and Professor Cope; to those of the Academy of Natural Scienceof Philadelphia, where are found other materials rendered precious by the laborsof the authors just mentioned; to those of the Carnegie Museum, Pittsburgh;the Field Natural History Museum, Chicago; the University of Kansas, Lawrence;the University of Texas, and the University of Nebraska. At Yale Universitythe writer has been permitted to study and describe valuable materials broughttogether by Professor O. C. Marsh, besides other specimens which form thetypes of species described more recently by Dr. George R. Wieland. Thru thecourtesy of Professor W. S. Valiant some of Dr. Leidy 's types preserved at RutgersCollege were made accessible. Specimens for study have been sent to the writerfrom most of the museums mentioned; also from the University of Chicago, byMr. S. W. Williston; from the University of California, by Dr. J. C. Merriam;from the Geological Survey of Canada, by Mr. L. M. Lambe; and from theVanderbilt University, Nashville, by Professor L. C. Glenn.

It has been the author's earnest wish to see all the types of the hitherto described

species; and most of these have come under his notice. Unfortunately some arewithout doubt utterly lost; others have, for the time, disappeared from view.

The author has endeavored to illustrate as fully as possible the species described.Whatever mayprove to be the little or greatvalue of the text, the writer can commendthe illustrations. By far the greater number of the figures of the plates are fromphotographs which were taken at the American Museum of Natural History byMr. A. E. Anderson. Other photographs and drawings have been furnished bythe Geological Survey of Canada thru Mr. L. M. Lambe; others by Dr. George R.Wieland, of the Yale University Museum; others by Dr. W. J. Holland, directorof the Carnegie Museum, Pittsburgh. The photographs of Basilemys sinuosacame from the Field Natural History Museum, thru Dr. E. R. Riggs. A consider-able number of the drawings found in the plates were prepared many years ago,for the late Dr. George Baur, by the United States Geological Survey; thru thecourtesy of the Survey these were placed in the hands of the writer. The major-ity of the drawings that appear in the text were executed by Mrs. Lindsay Mor-ris Sterling, artist in the department of vertebrate paleontology in the AmericanMuseum of Natural History. A number of these text-figures are the work of Mr.

Ill

IV PREFACE.

R. Weber, of the same museum. Most of the wash-drawings that appear in theplates were produced by Mr. Erwin C. Christman of the American Museum ofNatural History.

The thanks of the author are due, first of all, to the Carnegie Institution ofWashington for the support it has given him in the preparation of this monograph;in the next place, to Professor H. F. Osborn, of the American Museum of NaturalHistory, New York, for the opportunity to use the materials in that museum;finally, to all the persons and institutions above mentioned and others who havecontributed in any way to whatever there may be of value in this work.

It is not believed to be necessary to append a list of the papers and memoirsconsulted in the preparation of this monograph; inasmuch as the writer's Bibliog-raphy and Catalogue of the Fossil Vertebrata of North America, Bulletin 179 ofthe United States Geological Survey, records the literature of the subject up to theyear 1901 and the important writings since published are cited in the text of thismonograph.

Washington, November 20, 1907.

THE FOSSIL TURTLES OF NORTH AMERICA.

ON THE SOURCES OF OUR KNOWLEDGE OF THE FOSSIL TURTLES OFNORTH AMERICA.

The study of the fossil turtles of North America began in 1851, when Dr.Joseph Leidy described, under the names Stylemys nebrascensis, Testudo lata,Emys hemispherica, and Emys oweni, the extremely common fossil turtle of theOligocene Badlands of the present state of South Dakota, then a part of the Terri-

tory of Nebraska. It is true that in 1842 Dr. Richard Harlan had described andfigured (Amer. Jour. Sci., xliii, p. 144, pi. iii, figs. 2,3) as Chelonia coiiperi a bone

which he believed to be the femur of a large sea-turtle; but it is quite certain thatthe bone was no part of any turtle. Dr. Leidy continued at intervals up to 1889 todescribe new species of fossil turtles. Most of Dr. Leidy's shorter papers appearedin the Proceedings of the Philadelphia Academy, but his most important illustratedworks on the subject are found in D. D. Owen's Geological Survey of Wisconsin,etc., 1852; in the sixth volume of the Smithsonian Contributions to Knowledge,

1854; and in the first volume of the monographs of Hayden's Geological Surveyof the Territories, 1872. Professor E. D. Cope began his publications of speciesof fossil turtles in 1867; his last paper containing matter on the subject was pub-lisht in 1899, after his death. His papers on the subject are numerous and willbe found cited in the writer's Bibliography and Catalogue of the Fossil Vertebrata

of North America, 1902. Cope's most important expositions of the turtles are to

be found in the fourteenth volume of the Transactions of the American PhilosophicalSociety, 1869 and 1870; in the second volume of the Hayden Survey monographs,

1875; in volume four of Wheeler's Survey West of the looth Meridian, 1877; andin the third volume of the monographs of the Hayden Survey, 1884.

Professor O. C. Marsh described only three species of fossil turtles. No newfossil species are to be credited to Dr. George Baur, but he was the author of anumber of papers which made important additions to our knowledge of theirstructure and relationships. Dr. S. W. Williston and Dr. E. C. Case have con-tributed a number of valuable papers on the subject, especially on the turtlesderived from the Niobrara beds of Kansas. Dr. George R. Wieland has given

especial attention to the marine turtles of the Upper Cretaceous; and, especially,he has described that remarkable chelonian monster Archelon ischyros.

Other authors who have busied themselves more or less with the North Americanchelonian population of past times are J. Z. Gilbert, L. M. Lambe of theCanadian Geological Survey, Dr. F. B. Loomis, W. J. Sinclair, E. S. Riggs, andO. P. Hay. The results of the studies of the authors mentioned are that more than260 species of fossil turtles are now known from the geological deposits of NorthAmerica; the nearly complete structure of a considerable number of these hasbeen determined and much of that of others; and much light has been gainedregarding the history and relationships of the members of the order. Undoubtedlymany additional species will be discovered as the years pass away and species nowknown only from a few bones will become far better known.

I

4 FOSSIL TURTLES OF NORTH AMERICA.

costals as rib-heads, which articulate with the vertebral centra. At the front of the

shell each rib-head articulates with two centra, at their junction with each other;

more posteriorly, the rib-head falls farther and farther behind the intervertebral

articulation. There is no movement of the rib-heads on the centra, nor of these onone another.

The first dorsal vertebra articulates with the last cervical, permitting a veryfree motion. Its ribs are short, have no costal plate, and lie closely joined to the

front of the rib of the first costal plate. From this it will be seen that the secondrib has coalesct with the first costal plate, the third rib with the second costal, and

so on. The tenth dorsal rib is short and slender; its distal end is consolidated withthe eighth costal plate, and has the upper end of the ilium abutting against it.

In life the bones of the carapace and the plastron are covered by a number ofhorny plates, the scutes. Where the edges of any two of these meet they impress afurrow in the bone, forming a sulcus. When the skeleton is prepared the scutesfall off, but the sulci remain to betray the number and form of the scutes. In thefigures here shown (figs. I, 2), as in nearly all the figures of this work, the sulci arerepresented by stippled bands; the sutures between the bones, by zigzag Hnes.In fig. I the scutes are represented on both sides of the shell, but the symbols of

the names are placed only on the right side. Immediately in front, at the midHne,there is a small scute, the nuchal {nu. s). Then comes a row of five large scutes,the vertebrals {y. i,v. 2, etc.), each extending out beyond the neural bones. Thesulci separating these scutes cross respectively the first, third, fifth, and eighth neu-rals. On each side of these vertebrals is a row of four large scutes, the costals (c. s,i; c. s, 2; etc.). The sulci between these descend respectively on the second,fourth, sixth, and eighth costal bones.

The borders of the carapace are invested by a series of marginal scutes, twelveon each side (i, 2, 3, etc., on right side). The sulci dividing these from the costalscutes run along near the upper border of the peripheral bones. At the free bordersof these peripherals the scutes turn down and appear on the under sides of thebones.

It will be observed that the scutes coincide neither in number nor position withthe underlying bones. It is seldom that the sulci follow the sutures. This matterwill be discust hereafter.

In fig. 2 are represented the plastral scutes, the characters indicating the namesappearing in the right side of the drawing. A median longitudinal sulcus runsfrom the front to the rear of the plastron, separating the scutes of each pair. Infront is a pair of gular scutes (g); then a pair of humerals {hum); followed by thepectorals {pec), the abdominals {ab), the femorals {fern), and finally, by the anals{an). Just behind each axillary notch is an axillary scute; while just in front ofeach inguinal notch is found an inguinal scute. The marginal scutes seen in thisfigure are the same that appear in fig. i.

We may now examine the vertebrae in front of and behind the dorsals. In allof the turtles there are normally 18 presacral vertebrae, of which 8 belong to theneck. The more anterior and the more posterior cervicals are the shorter. Theneck as a whole is about as long as the dorsal series of vertebrae. The first is com-posite, consisting of four distinct pieces. On each side is a neural arch, aiding informing the neural canal. Below, these abut on a median piece, the hypocentrum.These three bones unite in forming a concavity, into which fits the ball-like occipitalcondyle. Behind the arches and the hypocentrum is the odontoid process, the propercentrum of the first cervical. Behind, this articulates with the centrum of thesecond cervical but does not become anchylosed with it.

OSTEOLOGY. 5

Of the other cervicals the arches are separable from the centra along a line ofcartilage. The neural arches are all low and devoid of spines. There are nolateral processes. As regards the articular ends of the centra, there is great diversity.

In Trachemys scripta, the second and third are convexo-concave; the fourth and

the eighth are convexo-convex; the fifth is concavo-convex; and the seventh, concavo-

concave. The articulations between the fifth and sixth, the sixth and the seventh,and the seventh and the eighth are extended from side to side and divided medially,

forming a true hinge joint, which permits very free motion up and down, but

restricts it from side to side. The anterior and the posterior zygapophyses arelittle elevated above the centra and are placed far apart. These arrangements

contribute to free motion in a perpendicular plane, but limit it in a horizontal.

Behind the last dorsal vertebra come two sacrals. Their ribs expand distally

and articulate with the upper ends of the iha. The sacrals are followed by a numberof caudals, about 15, but varying with the species, or even in individuals. Most

of these have transverse processes.

The skull of Trachemys scripta seen from above (fig. 3) presents three pairs ofbones which join at the median line. In front are the prefrontals, extending back-

ward to the middle of the orbits. The anterior ends roof over the nasal cavity.A strong process descends from each to join the vomer and the palatine, and toform the front wall of the orbit. Behind the prefrontals are placed the larger

jrontals. They aid in forming the rim of the orbits. The parietals are importantbones, inasmuch as they form the roof and much of the lateral walls of the brain-case. The anterior end of each sends downward a strong process which joins thepterygoid. Besides the pterygoid, the lower border of the parietal articulates with

the prootic and the supraoccipital. The latter bone is greatly prolonged backward,as it is in almost all turtles. It forms a small part of the boundary of the foramen

magnum.The maxilla bounds the nasal cavity on the side, the orbit below, and its lower

border forms an acute cutting-edge. In life this edge is covered with a horny

sheath. Posteriorly the maxilla articulates with the jugal. The hinder part of therim of the orbit is formed of the jugal below and of the postfrontal above. These

two bones form a postorbital bar of moderate width.

A large tympanic cavity is excavated in the quadrate, an extremely importantbone among the reptiles. Below the cavity mentioned, the quadrate descends toform a movable articulation with the lower jaw. In the hinder border of the quad-

rate is a small, but deep notch for the passage of a long, rod-like bone, the columella.

Interposed between the anterior border of the quadrate and the jugal bone is the

quadratojugal. The jugal and quadratojugal form the zygomatic arch. Aboveand behind the quadrate is the squamosal.

The sides of the hinder part of the skull are occupied each by a long excavation,the temporal fossa. The floor of this is formed of the parietal on the inside, of theprootic and paroccipital in the middle, and of the quadrate and squamosal on the

outside.

Fig. 4 represents the skull of the same species as seen from below. In front arethe premaxillce, bounding the nasal cavity below and entering into the roof of the

mouth. On each side and behind the premaxilla is the maxilla. It presents out-wardly the cutting-edge already mentioned. Its inner border joins, in front, the

vomer, posteriorly the palatine; the middle portion is mostly a free edge, forming

the lateral boundary of the choana. Between the two borders is a broad triturating,

or alveolar, surface, which in life is covered with a horny sheath. Along the middle

of this, parallel with the cutting-edge, is a sharp ridge, slightly tootht.

FOSSIL TURTLES OF NORTH AMERICA.

The midline behind the premaxillje is occupied by the single vomer. Anteriorly

it divides the nasal passages from each other; laterally it articulates with the

palatines; posteriorly, with the pterygoids. The palatines assist in roofing thenasal passages and in forming the triturating surface mentioned. Between each

and the maxilla of its side is seen an opening, the posterior palatine foramen.

The pterygoids meet each other at the midline anteriorly, but posteriorly are

separated by the basisphenoid. They extend backward so far as to exclude thebone last mentioned from contact with the quadrates. The lateral border of eachpterygoid is mostly a sharp free edge. Behind the basisphenoid comes the hasiocctpi-

tal. It is joined on each side by the exoccipital, and all three of these bones join in

forming the occipital condyle. From this view is seen also a portion of the par-occipital and squamosal. On each side of the basicranial axis are seen foraminafor the passage of various nerves and blood-vessels.

Each ramus of the lower jaw is composed of six bones. In front is the dentary,furnishing the triturating surface of the jaw, covered above with a horny sheath, and

pnuc

Figs. 3"andj4. Trachemyfscripta.

3. Skull seen from above. Xi. /r, frontal; _/u, jugal; /iii, parietal; /)aoc, paroccipital; />/r, prefrontal; ^0/, post-

frontal; pro, prootic; j«, quadrate; jy, squamosal; 50c, supraoccipital.

4. Skull seen from below. Xi. a/r, alveolar surface of maxilla; 6of, basioccipital; ^jp, basisphenoid; «.voc, exoc-cipital; mXy maxilla; paly palatine; paoCy paroccipital; pmx, premaxilla; pro^ prootic; pty pterygoid; jw. art,articulation of quadrate with lower jaw; gy, quadratojugal; 55, squamosal; voniy vomer.

completely co-ossified with its fellow of the opposite side at the symphysis. On thelower border of the jaw this bone extends backward nearly to the articulation withthe quadrate. The upper border of the jaw, behind the triturating surface, is formedin front by the coronoid bone; posteriorly by the supraangular. These two bonesare to be seen both from the outside and from the inside of the jaw. Behind thesupraangular is a nodular bone that articulates with the quadrate, the articular.On the inner surface of the jaw, near the hinder end, are two bones, whose namesare in dispute. Baur (Anatomischer Anzeiger, xi, 1896, p. 413) calls the lower ofthe two the splenial, the upper the angular. Williston (Science, xviii, 1903, p. 830)regards the lower bone as the angular, the upper as a dermal articular. The sameauthor has, in his work on North American Plesiosaurs, 1903, page 30, called thelatter bone the prearticular, and this name is adopted by the present writer.

The shoulder-girdle of an emyd turtle consists of two bones on each side. Oneof these, the scapula, consists of two slender portions placed at nearly a right anglewith each other. The longest portion, the proper scapula, the body of the scapula,

OSTEOLOGY. 7

extends from the glenoid cavity, of which it furnishes more than half, upward andinward, to become ligamentously attacht to the upper end of the first costal plate.The other part of this bone extends from above the glenoid cavity inward andforward, to be attacht by ligament to the entoplastron. This process has been

variously interpreted, but here it is regarded as the procoracoid, which has becomeco-ossified with the scapula. It will be called the procoracoid process. For a dis-cussion of this subject and a list of writers who have considered it the reader isreferred to a paper by Max Fiirbringer in Jenaische Zeitschrift, xxxiv, 1900.The other bone of the girdle is the coracoid. It starts from the glenoid cavity,and proceeds inward and slightly backward, to approach closely its fellow at themidline.

The writer regards the chelonian limb as belonging to a relatively primitivetype. If the reader will peruse Huxley's chapter on the position of the limbs, in

his Anatomy of Vertebrated Animals (Appleton's edition, 1872, p. 33), and com-pare his statements with what he can see in the limbs of a turtle, he will probablyagree with the view here presented. The apex of the angle at the elbow, instead ofbeing directed backward, is rather directed forward and upward. There is nocrossing of the ulna and radius.

The humerus of the emyd is rather strongly bent in the plane passing throughthe axes of the three segments of the Hmb. The head is directed upward. Thestrongly developt tuberosities for attachment of muscles, the radial, or lateral, andthe ulnar, or medial, are bent toward the lower face of the bone. The ulnar isalways the larger. At the distal end of the bone, on the anterior or radial side, is apassage for the radial nerve, the ectepicondylar foramen.

Of the two bones of the lower arm the radius is the smaller. It has a shallowcavity at the proximal end that meets the articular end of the humerus. Its distal

end is expanded and articulates with the intermedium and the radiocentrale of thecarpus. The ulna is the stouter bone, is flattened, has the suggestion of an olecranonprocess, and articulates distally with the intermedium and the ulnare.

The carpus is simple. Besides the bones already mentioned as belonging to thecarpus, there is another in the proximal row, on the outer side of the ulna. Thereare 5 bones in the distal row, but sometimes the fourth and the fifth are co-ossified.There are 5 metacarpals, each of medium length. The digits are 5 in number, thefirst and the fifth the shortest. The first digit has 2 phalanges, the others 3 each,exactly as in the Mammalia.

The pelvis is broad and short, and each half is composed of 3 bones, all ofwhich take part in the formation of the acetabulum. The ilium is expanded antero-posteriorly above and is articulated with the outer ends of the sacral ribs. Themiddle of the bone is slender. Each ischium has a posterior process which restson the xiphiplastron and an anterior process which runs forward to join the pubis

of its side. There is thus produced on each side a large heart-shaped fontanel,the ischio-pubic foramen. Each pubis has a lateral process which rests on thexiphiplastron. The pubis is prolonged forward considerably, and between the twothere is a median notch which in life is filled with cartilage, the prepubic; but thisdoes not become ossified.

The femur resembles considerably the humerus; but it is a longer bone andthere is no perforation corresponding to the ectepicondylar foramen. The head islarger and of different form. The trochanters are of about the same size and thedigital fossa separates them far down. The tihia is a stouter bone than the fibula.Its upper end is the larger and presents a large surface for articulation with the femur.

The lower end has a saddle-shaped articular surface to join the large bone forming

8 FOSSIL TURTLES OF NORTH AMERICA.

the first row of the tarsus. The fibula is slender, with the broader end downward,articulating with l^he same bone as does the tibia. The bone of the tarsus justmentioned is the only one present in the upper row, and is regarded as representing

three bones that theoretically belong in the upper row, besides the centrale. In the

lower row of tarsal bones there are five present, all articulating above with the

large bone of the upper row and distally each with one of the metatarsals. Thefifth of these bones is peculiarly expanded in the turtles.

The 5 metatarsals are rather elongated in Graptemys and Trachemys, as are,too, the phalanges. In the first digit there are 2 phalanges; in each of the others, 3.All the terminal phalanges are invested in horny claws, except the fifth.

THE CHELONIID/E. THE SEA-TURTLES.

Some of the salient features of the Cheloniidae will now be described. Theseconstitute a family of the Cryptodira, a superfamily to which also the Emydidaebelong. The Cheloniidae are greatly different in some respects from the Emydidae.The shell is composed of the same elements, but many are less completely developt.In general, the border of the carapace is excavated in front for the neck; and oneach side of this for free movement of the fore limbs. Behind, it is pointed, sothat in form the carapace is somewhat heart-shaped. The costal plates fall shortof reaching the distal ends of the ribs, as a result of which open spaces, or fontanels,are left between the costals and the peripherals. Usually none of the costal platescome into contact with the peripherals. The ribs of the costals, however, reachthese bones and each enters a pit in a corresponding peripheral. In the loggerhead(Caretta caretta), for example, rib-ends enter peripherals 3 to 8 inclusive andperipherals 10 and 11. In some forms, as Lepidochelys, there are supernumerarybones interposed among the neurals, so that the number of these seems to be ashigh as 14.

The carapace is furnisht with scutes similar to those of the Emydidae. InCaretta and Lepidochelys there is a supernumerary costal scute in front of thenormal first.

Altho the plastron of these turtles is composed of the same elements as that ofthe Emydidae, some of these are greatly modified. The hyoplastra and the hypoplas-tra are not suturally connected with the peripherals of the bridge; nor do theplastral bones just named come into contact with their fellows at the midline, sothat the space between the bones of the right and of the left sides is occupied by agreat fontanel. On each side there is another fontanel inclosed by the bridgeperipherals, the outer end of the hyoplastron, and that of the hypoplastron. Theinner and the outer borders of the two sets of bones last mentioned send out anumber of digitations into the fontanels. The epiplastra are saber-shaped boneswhich join each other and the entoplastron at the midhne, without jagged sutures;and their distal ends are applied to the outer border of the corresponding hyo-plastrals. The xiphiplastra are narrow, curved bones that approach each otherat their distal ends.

Carapaces and plastra similar to those of the Cheloniidae are found among theThalassemydidas, and to these the reader is referred for illustrations.

The neck of the Cheloniidae is much shorter than that of the Emydidae and thehead can hardly be retracted within the shell. In the loggerhead the neck is butlittle more than half as long as the series of dorsal vertebrae. The eighth cervicalarticulates by a synovial surface with the inferior side of the nuchal bone. Theseries of caudal vertebrae is short.

OSTEOLOGY. 9

The skull of the Cheloniidae differs in some important respects from that of theEmydidae. In order to illustrate this, figures are here introduced giving upper

(plate I, fig. i), lower (plate i, fig. 2), and lateral (plate 2, fig. i) views of the skull

oi Leptdochelys kempt Garman. These figures were prepared for Dr. George Baur,in 1888, but reverted to the United States Geological Survey, the then director of

which, Dr. C. D. Walcott, has permitted them to be used here. No figures ofthis species have hitherto been pubUsht. The most interesting feature of theskulls of the sea-turtles is the great extent of the bony roof covering the temporal

region. This roof extends from the orbit to behind the plane of the occipital con-

dyle. The postfrontal bone, narrow in the emyds, is carried backward nearly tothe hinder border of the roof. The squamosal sends upward and inward a platethat meets a horizontal plate from the parietal, forming a parieto-squamosal arch.

1 he lower side of the skull is interestingchiefly because of the broadening of the triturat-

ing surfaces of the jaws. The palatal plates ofthe vomer extend backward until they meetsimilar plates from the palatines. The choanaeare thus thrown much farther toward the middleof the skull than in the skull represented by text-

figure 4. The crushing surfaces of the lowerjaw are correspondingly widened (plate I, figs.

3 and 4). In the Thalassemydidae the choanaeFig. s.-Care^tta^caretta. Pelvis from ^^^ ^^ ^^^^^ ^^^^ ^^^^^^^ backward, aS may

be seen by examining the skull of Rhetechelys

platyops (Cope). There are no posterior palatine foramina.In the Cheloniidae the procoracoid process makes an obtuse angle with the body

of the scapula, and there is a distinct neck between the process and the glenoidfossa. The coracoid bone is longer than the scapula and moderately expanded atits free end. These bones are represented by fig. 2 of plate 2.

The humerus of these sea-turtles is strongly modified from the primitive form.It is much straighter than that of the Emydidae and has become flattened in theplane of the distal end. The head and the radial and ulnar tuberosities have changedpositions. The head is turned downward and proximad into the plane of the bone,while the tuberosities are lifted into this plane, each on its proper border of the bone.The radial tuberosity is carried along on the shaft of the bone until it has lost itsconnection with the head. It becomes divided into two parts, one for the deltoidmuscle, the other for the supracoracoid. Fig. 3, plate 2, represents the humerus ofLepidochelys as seen from above; fig. 4, as seen from below.

Relatively to the humerus, the radius and the ulna are shortened, the ulna

more than the radius. In the carpus the radiale remains small, the intermediumand the ulnare are enlarged and flattened, while the centrale is distinct. There arefive bones in the distal row, one articulating with each metacarpal. The third andthe fourth may be co-ossified. On the ulnar side of the carpus there is a large flatbone assisting to broaden the carpus. It may be regarded as the pisiform.

The second, third, and fourth digits are greatly elongated, the bones flat-tened, and all are bound together in a mass of muscles and skin to form an undividedoar. The first, and in some cases the second, digit is provided with a claw.

The pelvis of the Cheloniidae is broad and deprest (fig. 5). The ilia are short;the upper end is slender and turned backward and the axis of the bone is nearlyin the plane of the pubes. The latter bones are broad. Between the two, at the

10 FOSSIL TURTLES OF NORTH AMERICA.

midline in front, there is a deep notch, filled in life by the prepubic cartilage. The

lateral processes are broad. The ischia are without posterior lateral processes, and

they are not connected with the pubes on the midline by bone; thus the ischio-

pubic foramina unite into one in the prepared skeleton.

The femur, which in the emyds is longer than the humerus, is, in the sea-turtles,

shorter than the latter element. The shaft is not so straight as is that of the humerusand the head is not so much deflected toward the axis of the shaft. The trochantersdiffer considerably in size and the digital fossa does not descend far between them.

The tibia and the fibula are relatively stout bones. While in the emyds describedabove, the two upper segments of the fore limb are considerably shorter than the

corresponding segments of the hinder, the reverse is true in the Cheloniidae. Figs.

5 and 6 of plate 2 represent the femur of Lepidochelys.

The hinder foot of sea-turtles is, compared with the fore foot, greatly reduced.In Caretta the hinder foot, including the tarsus, is only about half as long as the

fore foot. There are 2 bones in the first row of the tarsus, 5 in the second row.

THE TRIONYCHOIDEA. THE SOFT-SHELLED TURTLES.

We will give our attention now to members of another superfamily, the Tri-onychoidea, popularly known as soft-shelled turtles, sometimes as mud-turtles, orriver turtles. The two last-mentioned names are, however, often applied to othergroups of turtles.

In these, as in other turtles, there is a carapace and a plastron; but they differ

greatly in appearance from those of the emyds. The whole body is greatly deprest.For illustrations of these portions of the shell the reader may consult the figures ofthe fossil species on succeeding pages. The carapace is usually composed of neurals,7 or 8 in number, 7 or 8 pairs of costals, and a nuchal bone. Only in one genus,Trionyx (EmyJa Gray), including two Asiatic species, are there any suggestions ofperipherals. The costal bones are closely articulated to their neighbors and thecontiguous neurals; but often these costals do not extend to the ends of the ribs;

or, if they do, it is only in very advanced age. In Hfe the periphery of the shell is

surrounded by a rim of dense and flexible connective tissue, terminating all round

in a sharp edge. It is this rim that has suggested the name soft-shell.The whole upper surface of the carapace is ornamented with a network of

ridges which inclose pits of various forms. The size and arrangement of the ridgesand pits vary in the different species. There are never any traces of horny scutes,the outer skin always remaining soft.

The plastron in some respects resembles that of the sea-turtles, inasmuch asthere are median and lateral fontanels and the connection with the carapace isa ligamentous one. The hyoplastra and the hypoplastra are furnisht with medianand lateral digitations. Of the median digitations one belongs to each hyoplastronand is directed forward and inward, to come into contact with the entoplastron.Two belong to each hypoplastron, one being directed toward that of the oppositeside, the other toward the xiphiplastron. Of the lateral digitations one set isdirected from each hyoplastron forward and outward, the other from the hypo-plastron backward and outward. The hyoplastron and the hypoplastron of eachside are closely sutured together, in some cases co-ossified.

The entoplastron is V-shaped, the apex being directed forward, an arm restingagainst the hyoplastron of each side. The epiplastral bones are slender and curved,usually not in contact with each other at the midhne. They differ from thesebones in all other turtles in being excluded from contact with the hyoplastra bythe arms of the entoplastron.

OSTEOLOGY. I

I

The xiphiplastra are usually more or less curved bones, whose anterior endsinterdigitate with the hypoplastra, while their posterior ends meet at the midhne.

In some cases the plastral bones are smooth and devoid of ornamentation. Inmany other cases there are developt on their lower surfaces patches of more super-ficial bone, the callosities. These callosities are sculptured into pits and ridges,somewhat like the bones of the carapace, but the pattern may be different. Eachcallosity may occupy but a small part of the bone developing it or it may extendover the whole bone. The callosities are more extensive in the Cretaceous andTertiary species than in those now existing.

The best description of the plastral bones of living Trionychidae has beenpublisht by Dr. Friedrich Siebenrock (Sitzungsber. Akad. Wiss. Wien, cxi, 1902,

pp. 807-846).

In the Plastomenidae the median fontanels are filled up; and the whole lowersurface of the hyoplastra, hypoplastra, and the xiphiplastra is sculptured.

The neck of the soft-shelled turtles is long and slender, and, hke that of theCryptodira, it bends most freely in a perpendicular plane. In a specimen of Platy-peltis spinifera at hand all the centra are convex in front, concave behind, exceptthat the hinder end of the eighth is reduced to a thin edge. This cervical is con-

nected with the first dorsal almost wholly by the zygapophyses.The skull of the soft-shelled turtles presents many distinctive features. Fig.

I, plate 3, represents that oi Platypeltis ferox, a Florida species, seen from above;fig. 2, plate 3, the same skull as seen from below. The skull is elongated and pointedin front. The posterior region is notable for the three large backwardly directedprocesses, the supraoccipital and the two squamosal processes. Of the temporalroof there is no part except the postorbital and the zygomatic bars. The prooticextends much in front of the articulation of the lower jaw. The hinder border ofthe pedicel of the quadrate is closed behind the stapes, so that this occupies a canal,

instead of a notch. The premaxillae have coalesct into a single small bone. Seenfrom below, the skull presents in front a median prepalatine foramen and the twochoanae. Altho the latter open rather far backward, they are not underfloored by

vomerine, palatine, and maxillary plates. The triturating surfaces of this speciesare rather broad. The vomer is always small. The palatines meet throughouttheir length at the midline, and posteriorly they articulate with the basisphenoid.

The pterygoids are broad, run backward alongside the basioccipital, and do notmeet on the median line. The basioccipital and the exoccipitals all take part inthe occipital condyle. The paroccipital is a large bone.

The lower jaw (plate i, fig. 5) is composed of the same elements as that of theemyds. The coronoid bone is large and the angle of the jaw projects considerablybehind the articulation with the quadrate.

The hyoid apparatus is strongly developt.In order to accommodate itself to the flattened form of the body, the scapula is

directed from the glenoid fossa upward, forward, and inward, making an angle ofabout 50 degrees with the procoracoid process (plate 3, fig. 3). The latter is slightlyexpanded toward its distal end and flattened. The coracoid is broad and flat andsomewhat saber-shaped.

The humerus (plate 3, figs. 4, 5) is a stouter bone than that of the emyds abovedescribed, and it is less neatly modeled. The proximal tuberosities, the radial andthe ulnar, are larger, with a broader fossa between them. The ectepicondylar passageis a groove. The radius is half the length of the humerus; the ulna is still shorter.

The fore foot is as long as the humerus, a condition due to the elongation ofespecially the median digits. The three on the radial side have claws, the others

12 FOSSIL TURTLES OF NORTH AMERICA.

do not. The phalangeal formula of the first three digits is 2, 3, 3, as in Emydidae.The fourth may have 4, 5, or 6, while the fifth finger may have 3 or 4 phalanges.

The pelvis resembles much that of the Cheloniidae, but the opening representingthe united ischio-pubic foramina is larger. The ilium is short and slender andthe upper end is turned backward. The ischia have short posterior processes.

The femur (plate 3, figs. 6, 7) is slightly longer than the humerus, which itresembles considerably. It may be distinguisht by its having no ectepicondylargroove and by the narrower, more elongated, head. The trochanters are wideapart and wholly separated by the interdigital fossa.

The tibia is a stout bone nearly three-fourths as long as the femur. As in otherturtles, the tibia is a slenderer bone. The hinder foot is still more elongated thanthe anterior, that of P. spinifera being more than a third longer than the femur.The first three digits have the same number of phalanges as other turtles; that is2, 3, 3. The fourth digit may have 4 or 5 phalanges; the fifth, 2 or 3. The firstthree have claws.

THE PLEURODIRA. THE SNAKE-NECKT, OR SIDE-NECKT TURTLES.

It is necessary to make some observations on the osteology of the members ofanother superfamily, the Pleurodira. The shell is composed of the same elementsas in the emyds and often presents no important differences from that of the latter.There are genera in which the neurals are reduced in number and a few in whichthese bones have been wholly supprest. In a few living genera and in somethat are extinct there is present a pair of bones, unknown in the Cryptodira andTrionychoidea, the mesoplastrals, interposed between the hyoplastrals and thehypoplastrals. In Pelustos (Sternothcerus) these bones join across the plastron.In Pelomedusa they are small triangular bones occupying the middle of the bridgeonly. The same bones occur in the species of Ba'ena; and the reader may consultthe figures under that genus.

The shell of all Pleurodira differs from that of other turtles in forming suturalconnections with the pelvis. The eighth costal plates develop each a sutural surfacefor the upper end of the ilium. On each of the xiphiplastrals are two sutural scars,the anterior for union with the pubis, the posterior for union with the ischium.On other pages will be found figures of the species of Taphrosphys, which representthese articulations. In the Pleurodira there is probably always an intergular scutepresent. Sometimes the gulars meet in front of it.

The cervical vertebrae are, as in all turtles, 8 in number, but they differ greatlyfrom those of the Cryptodira and of the Trionychoidea. In contradistinction to theneck of the latter turtles, that of the Pleurodira is constructed for free flexure in ahorizontal plane. This is effected by having the centra joined by ball-and-socketjoints and by having the zygapophyses of the two sides placed close together andhigh above the centra. A description of these vertebrae is given from the neck ofHydromedusa tectifera.

The neck is nearly a third longer than the dorsal series of vertebrae. The firstcervical, unlike that of the other superfamilies, has all the elements consoHdatedand is two-thirds as long as the longest. As regards the articular ends of the centrawe have the following: The first and the seventh are concavo-concave; the second,third, and fourth are convexo-concave; the fifth and the eighth are convexo-convex;the sixth is concavo-convex. All the cervicals possess well-developt transverseprocesses, with broad bases. Along the lower side of the centrum of each runsa sharp crest. The postzygapophyses of the first two vertebrae are separated by

OSTEOLOGY. ^3

only a slight notch; all the others have coalesct. The prezygapophyses are closetogether, but have not coalesct. Those of the eighth are very small, thus greatlydifferent from those of the members of the other superfamilies.

In some of the other Pleurodira the intercentrum and the odontoid process ofthe first cervical are distinct from the arches. In the Pelomedusidae only the secondcervical is convexo-convex; all the others concavo-convex.

In none of the Pleurodira is the neck withdrawn into the shell as it is in the otherturtles, but is bent sideways and brought under the projecting borders of the shell.In harmony with this action, the anterior bones of the carapace and of the plastronoften project farther than in other turtles.

The skull of the Pleurodires offers many peculiar structures. That of Hydrome-dusa may first be considered. This skull is long and very flat. There are distinctnasal bones. The prefrontals do not send down processes to the vomer. There is apostorbital arch, but no zygomatic arch. At the rear there is a very slender palato-squamosal arch. Seen from above the whole upper surface of the pterygoids and thewhole temporal fossae are exposed to view. There is no ridge projecting outwardfrom the parietal over what may be called the suprapterygoid fossa, as in otherturtles; nor does the prootic project forward over this fossa. The fossa just men-tioned is bounded outwardly by the upturned outer border of the broad pterygoids.The premaxillas are small. The supraoccipital spine is extremely short.

Fig. 6 represents the skull seen from below. Thetriturating surfaces of the upper jaw are very narrow.The choanal openings are very large and are separatedby the splint-like vomer, which comes into contact with

the pterygoids. The latter bones separate widely thesmall palatines. Behind the latter bones are the posterior

palatine foramina. The pterygoids are greatly developtanteriorly and laterally. They are in contact on the mid-line in front, but for more than half their length thebasisphenoid bone comes between them. The pterygoidsare abbreviated behind, so that they permit the quadrates

to join the basisphenoid. This arrangement is character-

istic of the Pleurodires and distinguishes them from all

other turtles. The quadrate is notcht behind for thepassage of the stapes.

The lower jaw is slender and the coronoid processesare low. The articular furnishes a ball for articulationwith the quadrate. There is present what Baur called

a presplenial, but which is here regarded as the true

splenial, a bone absent from most turtles.

The hyoid apparatus is greatly developt, but need nothere be described.

The scapula is a strong bone. The procoracoidlation of quadrate with lower jaw; process makes Icss than a tight angle with the body ofvom, vomer. '

, , j-, ,~

. ~ . , . ^.the bone. Between the two portions, m the angle, is a

sharp crest. The glenoid fossa is at the end of a long neck. The coracoid isrelatively short, much bent in a horizontal plane, and expanded at the free end.

The humerus resembles that of the emyds in most respects. The ulnar andradial tuberosities are somewhat larger, the ulnar ascending slightly above the head,the radial descending lower than in the emyds. The ulna has no suggestion of an

Fig. 6.

—

Hydromedusa. Skull

from below. Xj.

bocy basioccipital; hsp, basisphenoid;

exoCj exoccipital; mr, rnaxilla; pa^parietal; pal, palatine; paoCy paroc-

cipital; pmXj premaxilla; />/, ptery-yoid; qu^ quadrate; qu. art, articu-

14 FOSSIL TURTLES OF NORTH AMERICA.

olecranon. The tarsus and digits do not differ enough from those of emyds torequire further notice here.

The pelvis of Hydromedusa differs greatly from that of the emyds and triony-chids. The ilia are much expanded and triangular at their upper ends, and havebecome joined by a rough suture with the eighth costals. The lateral processes ofthe pubes are stout and are sutured to rough surfaces on the xiphiplastra. Theanterior branch of the pubis is slender and joins its fellow at the midline. Theischia are stout bones, sutured to the hinder border of the xiphiplastra. The lowerportion of each ischium expands into a great lateral process, which runs forward

and inward to join the one from the other side, the whole length of each process

being sutured to the xiphiplastron. There is no bony union of the pubes and theischia along the median line. As shown by Baur (Jour. Morphology, iv, 1891, p.351) there is, in many Pleurodires, a greatly lengthened prepubic cartilage. It isto be compared with the prolongation forward seen in the pubes of species ofBa'ena and Chisternon.

The hinder limb does not differ notably from that of the emyds.Illustrations are here furnisht of the skull of Podocnetnis expanse, another

Pleurodire, the structure of which in many respects strikingly differs from that ofHydromedusa. From plate 4, fig. I, it will be seen that the temporal region isnearly as completely rooft over as is that of Lepidochelys (plate i, fig. i). Thereare no nasal bones and the supraoccipital spine is long. In one feature of thetemporal roof this turtle is different from Lepidochelys. In the latter the postfrontalbone extends backward nearly to the hinder border of the roof; in Podocnemisthe postfrontal bone is very small and its place in the roof is mostly occupied by thequadratojugal, which rises over the squamosal, excluding the latter from contactwith the parietal (plate 4, figs. 1-5).

Fig. 2, plate 4, represents a palatal view of the same skull. The premaxillae areof considerable size. The vomer is absent. The triturating surface of the maxillais broad and furnisht with ridges. The choanae are restricted. The palatines arebroad and meet throughout their length at the midline. The pterygoids, too, arebroad, little separated mesially by the basisphenoid, and the outer border of each isturned upward into a scroll-like process. The articular surface of the quadrate,for the lower jaw, is seen to be concave.

Fig. 5, plate 4, shows the same skull viewed from the side.The lower jaw (plate 4, figs. 3, 4) presents a ball-like articular surface for the

quadrate. In front of this are seen the angular, the prearticular, the splenial, andthe dentary, the latter consolidated with its fellow at the symphysis and furnishtabove with a broad and ridged triturating surface. On the outside of the hinder halfof the ramus is the supraangular.

THE DERMOCHELYID^E. THE LEATHERBACK TURTLES.

It is necessary now to describe in brief terms the skeleton of the leatherback(Dermochelys coriacea), one of the most extraordinary of turtles. It attains a greatsize and is the most thoroly aquatic turtle that is known (fig. 7). Figures of allportions of the skeleton may be found on the plates illustrating a paper by PaulGervais on this turtle, publisht in the Nouvelles Archives du Museum d'HistoireNaturelle de Paris, volume viii, 1872.

The ribs of this turtle, instead of being consolidated with costal plates thatunite edge to edge to form a carapace, are wholly free from one another. The firstrib is wholly free from the second, the tenth from the ninth. The only thing that

OSTEOLOGY. 15

may be supposed to represent the costal plates is the irregular border of the some-what expanded ribs. Of the other elements entering into the carapace of an emydor a pleurodirid, the leatherback possesses only the nuchal bone.

But the leatherback has a carapace peculiar to itself. This is composed of a

layer of thin, polygonal bones which are buried in the thick skin of the animal. Ofthese bones there are 7 rows of larger ones, that appear in the living animal as so

many sharp dorsal keels. One of these rows is along the midline; three run alongeach side. In front, this layer of mosaic-like bones overHes the nuchal. Smaller

bones fill up the spaces between the rows.

On the inferior side of the turtle there are 5 rows of similar bones, but the spacesbetween the rows are not so completely filled as on the upper side. Beneath the

skin which supports these rows of bones there is a ring of elongated bones which

represent the plastron of more normal turtles. These represent the epiplastra,

the hyoplastra, the hypoplastra, and the xiphiplastra. The entoplastron is missing.All these bones are slender and thin, and they surround a vast fontanel.

Fig. 7.

—

Dermochelys coriacea. Greatly reduced.

The cervical vertebrae differ in no important respect from those of Caretta, andthe neck is equally short. The dorsals are ten in number and immovably joined tothose in front and behind by rough articular ends. The neural arches are movedforward, so that each articulates about equally with its own centrum and that inadvance. They are somewhat expanded above, but do not come into contact withplates from the ribs. In fact, these plates are extremely vestigial. There are two

sacral vertebrae, whose ribs articulate with the iha; and there are about twenty

caudals.

The skull at first glance presents many resemblances to that of members of theCheloniidae. The temporal roof extends backward as far as the occipital condyle.The postfrontal and the jugal are large, and the squamosal joins the parietal.The supraoccipital spine is short. The prefrontals extend backward to beyond theorbits. The external nares look forward and strongly upward. The maxillae arenot strongly developt; and they have hardly any triturating surfaces. The choanaeare placed far forward and open directly into the roof of the mouth. A spHnt-likevomer separates them, and extends backward to the pterygoids. The palatines arebroad, and they reach forward nearly to the vomer, sometimes coming into contact

with it. Thus, they bound the choanae outwardly. The pterygoids join on the mid-line for a short distance only in front; otherwise, they are widely separated by the

l6 FOSSIL TURTLES OF NORTH AMERICA.

basisphenoid and the basioccipital; they shut off the latter bones from contact

with the quadrate. The inner borders of the pterygoids lie upon the upper side ofthe basisphenoid.

In one important respect the leatherback and its few fossil relatives differ from

all other turtles. This is in the fact that the parietal bone does not send downward,

in front of the exit of the trigeminal nerve, a plate to join the pterygoid.

The occipital condyle remains cartilaginous. The pedicel of the quadrate isslender and directed strongly forward. The lower jaw is feebly developt and fittedfor soft food. The articular remains cartilaginous. The angular and the supra-angular are short, as seen from the outer side of the jaw, but are prolonged on the

inner side. The prearticular and the coronoid are not present.The parts of the scapula are short and thick. The procoracoid process makes

an obtuse angle with the body of the bone. The coracoid bone is about twice aslong as the process mentioned, and the free end is somewhat expanded and flattened.

The humerus is a massive bone and much flattened. The general form is thatof the humerus of Leptdochelys, but the radial tuberosity is brought down to apoint more than half-way from the head of the bone to the distal end. The endfor articulation with the ulna and the radius is very broad, and nearly semicircularin outline. The ectepicondylar passage is a foramen wholly in the bone, about themiddle of its width.

The radius and the ulna are stout bones of about equal length and only abouthalf as long as the humerus. The carpus has 2 bones in the first row, a centrale,5 bones in the second row, and a large pisiform. Sometimes the fourth and thefifth distal carpals are co-ossified. The digits have the normal number of pha-langes; that is, 2 in the thumb and 3 in each of the others. The second, third andfourth digits are greatly elongated, being twice the length of the humerus measuredfrom the head. None of the digits bears a claw. All are bound together in a com-mon envelope of skin to form a powerful flipper.

The pelvis is in many respects different from that of the Cheloniidae. Theform of the ilium and its position with respect to the pubis do not differ greatlyfrom those of the ilium of Caretta, but the ischia are broader fore and aft than inCaretta. The pubes are connected along the midline for a much greater distancethan in the Cheloniidas. Posteriorly they approach the ischia, so as nearly to jointhem, the interspace being filled by cartilage. The ischio-pubic foramina areextremely small and nearly separated in the midHne. The lateral processes arevery large, extending outward and forward, so that at their extremities the pelvisis twice as wide as it is across the narrowest part of the pubes. Each processterminates in a large plate of cartilage. There is likewise a large spatulate prepubiccartilage.

The hinder limb is reduced in size, as in the Cheloniidae, and offers differencesonly in detail.

MODIFICATION IN TURTLES. 1/

ON THE AMOUNT OF MODIFICATION UNDERGONE BY TURTLESSINCE THEIR EARLIEST APPEARANCE.

On preceding pages a brief exposition has been presented of the principalstructures shown in the skeletons of turtles of various groups. It is now intendedto consider the differences of structure from another point of view, that of determin-

ing the amount of differentiation these animals have suffered since their earliestappearance, especially as compared with the changes undergone by other orders of

reptiles. It is generally supposed that turtles have been unusually conservative in

their changes; that, as in their movements so in their development, they haveprogrest slowly. We are to inquire to what extent this impression is true.

Let us first consider the shell, that part which is most often found preserved in

the rocks and the part which is regarded as most characteristic of turtles. Thetheory accepted by the writer is that originally the upper side of the body was

protected by a rim of peripheral bones, a median row of neural bones, and eight

pairs of bony plates which overlay the ribs and had possibly in the earliest turtle

coalesct with them, probably joining one another by their contiguous edges; and

that in addition to the bones enumerated, subdermal in their position, there was amore superficial layer of bones, dermal in position and forming seven longitudinal

zones, a median or vertebral zone, two costal, two supramarginal, and two marginal

zones. The plastron was formed of at least eleven subdermal bones, while super-ficially there were found five zones of dermal bones, a median and two lateral on

each side of it. Starting with this outfit, the vast majority of turtles have wholly

or almost wholly divested themselves of the dermal layers and have acquired a soHd

shell composed of the subdermal bones. On the other hand, the leatherbackturtle, Dermochelys, appears to have retained the dermal and to have almost wholly

surrendered the subdermal bones, for the costal plates now form only unimportantfringes on the ribs; the peripherals and neurals are wholly gone; the nuchal

is reduced; and the plastral bones are only eight slender rods. As regards the

carapace, Dermochelys possesses little that is homologous with that of most other

turtles.

If we view the modifications undergone by the shell within the group known asThecophora we find that they are extensive. We can hardly doubt that the primi-tive turtles possest a pair of mesoplastral bones; and yet we know that all turtleshave discarded these, except a few of the Pleurodira. Within the latter super-

family the shell has not suffered remarkable modifications. Nevertheless, most

of the genera have no mesoplastra. In some the plastron is connected with the

carapace ligamentously, while in others strong axillary and inguinal buttresses rise

from the plastron and articulate with the inferior surface of some of the costals. In

all, the hindermost costals have contracted a sutural union with the ilia. In a

considerable number of genera the neurals have become wholly supprest. In

Pelusios (Sternothcerus) there is a hinge behind the hyoplastra.

In the superfamily Trionychoidea the shell has undergone extensive reduction.

No traces are known of the original dermal layer of bones. The neurals and thecostal plates are retained; but the latter show retardation in their development,

attaining the distal ends of the ribs not at all or only at a late date in life. Theperipherals are wholly missing, except perhaps in the genus Trionyx {Emyda Gray).

In the plastron we never find mesoplastra and all the bones are more or less reduced.In the great majority of cases there are fontanels on the midline and at the bridges.

The form and relations of the entoplastron and the epiplastra are very differentfrom those of other turtles, but homologies are easily traced.

l8 FOSSIL TURTLES OF NOR! H AMERICA.

It is in the extensive group of Cryptodira that we meet with the greatest numberof what may be regarded as the minor modifications of the shell. If we exceptToxochelys, no known Cryptodire shows more than vestiges of the ancient layerof dermal bones. In some, as Protostega and Archelon, of the Upper Cretaceous,the costal plates are nearly as much reduced as in Dermochelys. In these and anumber of other genera of Cretaceous turtles the peripherals are slender, but theypersist. Probably the neurals are never wholly absent. The elements of theplastron degenerate in some cases; but, with the exception of the entoplastron of afew genera, all persist.

It is in this group that we discover the greatest variety in the forms of the shell,ranging in convexity from much deprest to highly vaulted and bombous, and fromrelatively long and narrow to a breadth greater than the length. There may beone or more carinae on the carapace and its free borders may be smooth or variouslynotcht or rolled.

In some genera of Cryptodira, as the snappers and the sea-turtles, there areextensive fontanels between the distal ends of the ribs and others at the sides andthe middle of the plastron. In most Cryptodira the bones are solidly articulated,abrogating all fontanels. In the snappers and the sea-turtles again, the plastronis only ligamentously joined to the carapace. In other genera, as Batagur, Hardella,and Echmatemys, the plastron sends up powerful axillary and inguinal buttresses tothe inside of the carapace. Between these extremes there are all gradations.

Among the Cryptodira there is a great variety of hinges between portions of theshell. In Cyclemys there is a hinge between the hyoplastron and hypoplastron,and both these bones are sutured to peripherals. In Terrapene there is a similarhinge, with the bones only ligamentously joined to the carapace. In the extinctgenus Ptychogaster there is a sliding joint between the hypoplastron and the contig-uous peripherals. In Kinosternon there is a hinge between the epiplastra and thehyoplastra, and another between the hypoplastra and the xiphiplastra. In Kinixysthere is a hinge in the carapace between the fourth and the fifth costal plates.

Attention may be called to the modifications of the neurals and the costal platesin Testudo. The neurals are alternately large and octagonal and small and quadrate.The costals are truncated wedges, placed so that broad and narrow ends alternateboth next the neurals and the peripherals.

There are numerous interesting modifications in the number, form, and disposi-tion of the horny scutes of the shell among the turtles. There are supposed to havebeen originally 12 rows, or zones, of these scutes, corresponding to the 12 rows ofdermal bones of Dermochelys and the ancestral turtle. A scute coincided with eachbone. In each row some scutes grew at the expense of the others and persisted evenafter the disappearance of its supporting bone. Whole rows of the scutes dis-appeared, as the supramarginals and the median plastral row of most turtles. Thesupramarginals are represented in Macrochelys by a few scutes over the bridges.The anterior and posterior ends of the supramarginal series are found in the TriassicProganochelys. Many genera furnish the inframarginals, as nearly all the Derma-temydidae. In the Emydida; all have vanisht except one in the axillary notch andanother in the inguinal. In the Baenidae, the Dermatemydidae, and the Pleurodirawe find the middle plastral row represented by the intergulars. Archa-ochelysLydekker and Polythorax Cope present more posterior scutes of this row. BetweenzyoungDermochelys with its 12 rows of epidermal scutes, with many in each row,and Terrapene, with only 7 rows, there are many and interesting stages.

But the Trionychidae carry the reduction of the scutes to the extreme, for none ofthe whole superfamily shows any traces of these whatever. In various species of the

MODIFICATION IN TURTLES. IQ

Other groups the boundaries of the scutes are not visible on the shells; but in mostof these cases the scutes were probably very thin and their edges did not impressthe bone. In the living Carettochelys and the extinct Pseudotrionyx, both Crypto-dires, there are no scutes.

The principal characters which belong to turtles in general will be given on asucceeding page. Besides these common characters, others that have been lost ormodified were possest by the more primitive members of the order. Of these maybe mentioned the presence of nasals, of lachrymals, and of an extensive roof over thetemporal region. The presence of this roof in the skulls of the older turtles is sogeneral, perhaps rather so universal, that its primitive nature can hardly be doubted.

Its frequent presence in the skulls of the lower forms of living turtles and its generalabsence in the case of the higher forms confirm this conclusion. This roof must havebeen inherited from the Cotylosaurian ancestors of the order. Otherwise, we mustsuppose that it was developt during the history of the chelonians and again lost bymost of them. It appears obvious that the roof has been reduced in proportion to theelongation of the neck and the ability to withdraw the head beneath the shell. InDermochelys and the Cheloniidae the roof extends backward as far as the occipitalcondyle; and all these turtles have short necks and can furnish beneath the shelllittle protection to the head. In the snappers the roof is moderately developt; in

most of the Emydidae and all the Trionychoidea the roof is reduced to a narrowpostorbital bar and to a zygomatic bar. In some species of Terrapene and in someother genera the zygomatic bar is missing. While in some Pleurodira, as Podocnemis,the roof is wide, in others nothing is left but the postorbital bar. Several genera

possess a parieto-squamosal arch (Rhtnemys, Hydraspis, Hydromedusa). Dr.George Baur showed the various ways in which these results have been attained.In the Cryptodira the roof has been eaten away from behind forward and in frontof the tympanic cavity, severing first the union of the parietal and the squamosal,then that of the postfrontal and the squamosal, and finally in rare cases that of thejugal and the quadratojugal. In the Pleurodira the reduction has followed thereverse course, beginning in front of the tympanic cavity and moving upward andbackward. In several genera of Pleurodira there has been left only a very narrow

parieto-squamosal bar; in others, not even this. In all cases the removal of the

bone has begun at one border or the other of the bone and never by the formation offontanels in the roof.

In no other order of reptiles do we find such enormous variation in the characterof the temporal roof. Superfamilies and families of turtles present modificationsthat would characterize orders of other reptiles.

The palatal region offers variations in structure that are of wide range. Primi-tively the choanae opened in the front of the mouth one on each side of the narrow andshallow vomer. A broadening of the triturating surfaces of the jaws required thatthe choanae be pusht backward. This was accomplisht by the development ofa median descending plate of the vomer, which spread laterally and joined palatalplates from the maxillae, the palatines, and sometimes even from the pterygoids.Thus a floor was formed beneath the nasal channels, carrying these backwardsometimes more than half-way to the occipital condyle. The process is similar towhat occurred in the order of Crocodilia; only, in the most primitive known formsof the latter the choanae are placed immediately in front of the pterygoids, while in

the most advanct forms the choanae are in the rear of these bones. These differenceswere regarded by Huxley as sufficient to justify the recognition of two groups ofcrocodiles, the Mesosuchia and the Eusuchia. Variationsof greater extent among theturtles characterize famiUes only or even genera.

20 FOSSIL TURTLES OF NORTH AMERICA.

The jaws have suffered extensive modifications. In the less differentiatedturtles the cutting-edges of the jaws are low and the triturating surfaces narrow.

The cutting-edges may become deep, without increase of the width of the trituratingsurfaces and vice versa. For an illustration of great width of the grinding surfaces

the reader is referred to the species Rhetechelys platyops (Cope).

The lower jaw becomes modified to correspond with the upper jaw. The ramiof that of Toxochelys latiremis are narrow and slender; while those oi Erquilinnesia

are broad, flat, and fitted for crushing hard objects.

The vomer, which is always developt in the Cryptodira, the Trionychoidea andsome of the Pleurodira, is wholly missing in some forms of the latter superfamily.

There is great variation in the pterygoids. The primitive condition appears tobe essentially that now found in the Cryptodires in which these bones are ofmoderate width and extend backward so as to exclude the quadrates from contactwith the basicranial bones. This condition exists also in the Amphichelydia andthe Trionychoidea. In the Pleurodira the pterygoids have become shortened poste-riorly, so as to let the basicranial bones join the quadrates. They have also becomeexpanded, and the outer edge is rolled up in a scroll-like manner.

The necks of turtles furnish us with many interesting features. There isno doubt that in the early turtles the neck was short, perhaps less than half as longas the dorsal series of vertebrae. Doubtless the lengthening has been broughtabout to facilitate the prehension of food, but it has had other consequences. Insome species of each of the superfamilies of Thecophora the neck is considerablylonger than the dorsal series, but it has retained its primitive shortness in Dermo-chelys and the other sea-turtles. The elongation of the neck is never due to anyincrease in the number of vertebras, but to the lengthening of the individual vertebrae.

The most important modifications of the cervical vertebrae are to be found, not intheir mere elongation, but in the structure of their parts. Originally, as we learnfrom Glyptops pUcatulus, the vertebral centra were all biconcave. In Ba'ena andChisternon we find the beginnings of differentiation in the articular ends of thecentra. The highest stage of differentiation is perhaps to be found in species ofTestudo. In Testudo radtata, of Madagascar, the second cervical is convexo-concave; the third and the eighth convexo-convex; the fourth, fifth, and sixth,concavo-convex; the seventh, concavo-concave. The joint between the sixth andseventh and that between the seventh and the eighth are elongated from side to sideand divided into right and left portions, forming true ginglymoid articulations.Variations of this arrangement are found even within the genus Testudo. Modifi-cations and less differentiated stages are to be found in other famiUes of Cryptodira.

The necks of the Trionychoidea are usually greatly elongated and the vertebraeare essentially as in the Cryptodira. The Pleurodira do not possess ginglymoidjoints, but in at least one genus there are saddle-shaped vertebral articulations, asin birds.

In the development of the mechanisms permitting flexion of the neck two distinctpaths were followed, the Athecae, the Cryptodira, and the Trionychoidea taking theone, the Pleurodira the other. The goal reacht in each case is very different fromthat attained in the other. In the first case, the neck is bent in a perpendicularplane ; in the case of the Pleurodira, in a horizontal plane. As the neck of membersof the Cryptodira and Trionychoidea lengthened, the head began to be withdrawnwithin the shell for protection. In the case of the Pleurodires the neck is bentlaterally, and it and the head are protected by the projecting borders of the shell.If the neck is so long that it would carry the head beyond the axilla it is bent firstin one direction, then in the other.

MODIFICATION IN TURTLES. 21

To permit these different modes of flexion the mechanical arrangements mustbe different. In the Cryptodires and the trionychids there are ginglymoid joints

at the base of the neck, facihtating bending in a perpendicular plane. The neuralarches are low, with the zygapophyses wide apart, favoring motion in the vertical

plane, restricting it in the horizontal. In the Pleurodires the arches are high, the

zygapophyses close together, often confluent, and at least one end of most of thecentra semiglobular, arrangements aiding motion in a horizontal plane.

It is doubtful whether there is another group of vertebrates that possesses so

many modifications of the cervical vertebrae as are found in the turtles.There has occurred a considerable amount of modification in the pelvis of these

reptiles. As shown by Glyptops, of the Jurassic, and Ba'ena and Chisternon, of theBridger, the pelvis was not originally suturally joined to the shell. In both theCryptodires and the trionychids the pelvis has retained its original freedom. Inthe Pleurodires the ilia effected, before the close of the Cretaceous, strong sutural

connections with the hindermost costal plates; while the ischia and the pubesbecame closely sutured with the xiphiplastra.

In the Amphichelydia the ischia and the pubes are joined along the midline bya bar of bone, thus defining right and left ischio-pubic foramina. In the Baenidae,so far as known, and possibly in Glyptops, the prepubic process was stronglyossified. In the Emydidje and the Testudinidae the ischio-pubic bar is ossified; butin the Cheloniidas and the Trionychoidea this region is wholly cartilaginous. Thatthe primitive condition of the ischio-pubic bar and of the prepubic process in theAmphibia and the early Reptilia was cartilaginous we can not doubt. That theseshould be ossified in the Amphichelydia is remarkable. It suggests that the carti-laginous condition in so many hving tortoises may be due to degeneration.

The limbs of most swamp-loving turtles present primitive conditions of thereptilian hmb, both with respect to their composition and their disposition. Thesegments of the Hmbs are mostly flext in one plane, and this plane stands more orless at right angles with the axis of the body. The apex of the angle at the elbowis directed forward, rather than backward as in the mammals, and the ulna and theradius do not cross. There have occurred few unions of bones and these are con-fined to the carpus and the tarsus. No bones have suffered important reductionsor modifications. In the hmbs of swamp-inhabiting turtles there have been fewchanges since Jurassic times; and, since the Amphichelydia, the Cryptodira, and

the Pleurodira possess similar limbs, it is evident that the primitive turtles possest

limbs not greatly different.

From this simple type of limb there has been divergence in two directions; oneto adapt the animal for life on the land, the other for habitual hfe in the water.

The highest expression of the former adaptation is perhaps to be found in the limbsof species of Testudo. In these the principal modification in the proximal bones

of the limbs is the drawing downward and toward each other of the tuberosities ofthe humerus. Most important is the shortening suffered by the bones of the digitsand the reduction of the number of phalanges in each to no more than two. Thefifth hinder digit may become vestigial. The result of these changes is the pro-duction of a short foot resembling that of an elephant and adapted for travel overrough and hard ground.

To fit the animal for habitual life in the water the anterior limb tends to beconverted into a flipper. First of all, the fingers become elongated to support abroad web. Usually the number of phalanges remains unaffected. In the Trio-nychidae some of the digits are much elongated and the phalanges are more numerousthan the normal, and two of the claws have disappeared. In the sea-turtles.

22 FOSSIL TURTLES OF NORTH AMERICA.

never coming on land except to deposit their eggs, the fore Hmbs have becometransformed into definite flippers with the fingers elongated and all bound togetherin a common mass of skin and muscles. Only one or two claws remain. Thephalanges and carpals are considerably flattened, as is also the humerus. Thelatter also becomes straighter and the insertions of the deltoid muscle descend onthe shaft. While these changes progrest the hinder limb became relatively smallerand fitted for steering the animal. The extreme of these modifications is to beseen in the limbs of the leatherback, Dermochelys. Between the short club-likefoot of Testudo and the long and powerful flipper of Chelonia and Dermochelys thereis a vast interval.

Amid all the changes that have occurred in the turtles certain fundamentalstructures have remained unafi^ected. The jaws have always retained their hornycovering. The quadrate has remained fixt. The cervical vertebra; have keptunchanged their number, 8, and the dorsals their original number, lo. All fourof the limbs have persisted and all the segments of each.

A consideration of the changes which turtles have undergone and a comparisonof these with the modifications suffered by other groups of reptiles lead to theconclusion that no other order of these animals, except the Squamata, can displaysuch a variety of structures. The Squamata, embracing the mosasaurs, the lizards,the chameleons, and snakes, have undoubtedly, in adaptation to different modesof life, diverged in more directions and gone farther than have the turtles. The skullhas become modified in more ways; the vertebrae, at least as regards the number inthe column, have varied more; the limbs have undergone more varied adaptationsfor walking, for climbing, for grasping, for swimming, for leaping; and in the wholegroup of snakes and in many lizards the limbs have wholly disappeared. To thesedenizens of the earth, swarming since probably the Triassic, the turtles must yield invariety of form and structure and habits, but probably to no other order of reptiles.

PRIMARY AND SECONDARY CHARACTERS. 27,

ON THE PRIMARY AND THE SECONDARY CHARACTERS OF TURTLES.

It may be interesting and productive of some useful result to endeavor to separatethe osteological characters that belonged to the most primitive turtles from thosethat may be called the secondary characters, those that have been acquired in latertimes through the exaggerated development or reduction or the suppression of cer-tain parts ol the skeleton or their modification of form and connections, in order toadapt them to new uses. We will, as heretofore, consider first the shell of the turtles.

As already stated, the writer holds the view that the earliest turtles possestpractically two kinds of shell, one purely dermal, consisting of probably a mosaicof small bones arranged in at least 12 longitudinal zones. Each zone probablyconsisted of a row of larger bones, bordered on each side by smaller ones. It is notnecessary to suppose that the spaces between the zones were wholly occupied bythese smaller bones. Each of these bones was covered by a horny scute. Thenearest approach to such a dermal shell is in our days seen in Dermochelys, as hasalready been stated.

Beneath the skin there seems to have existed a carapace more or less completewhich consisted of a nuchal, a median row of neurals, 8 pairs of costals, a pygal,probably one or more suprapygals, and about ii peripherals on each side. Towhat extent the neurals and the costal plates had become anchylosed to the neuralspines and the ribs respectively it is now impossible to determine. Nor can we sayto what extent the various elements of this carapace had become connected with oneanother. The existence of the dermal carapace would appear to indicate that thesubdermal box was not yet closed.*

There was a subdermal plastron that was composed of at least 11 bones.Portis has described Polysternon, which had an additional pair of bones between thehypoplastra and the xiphiplastra, making 13 all together. Between these variousbones there may have existed more or less extensive fontanels.

According to the author's views, as time went on the external, mosaic-like shelldisappeared in most turtles, while a more efficient armor was developt out of thesubdermal elements. In the ancestors of Dermochelys, however, the dermal armorwas retained, while the more deeply seated one disappeared, with the exception ofthe nuchal bone.

It is proper to state that all authors do not hold that Dermochelys has descendedto us in a direct line distinct from other sea-turtles, but has been derived from themat a more recent date. On this subject the reader must consult the papers of Baur,Boulenger, Case, DoUo, E. Fraas, Hay, Van Bemmelen, and Wieland.

Most Thecophore turtles have lost the bones of the outer dermal shell. Thereare yet traces of it perhaps in the dorsal and ventral keels of various turtles, in thetubercles that diversify these keels, and especially in the rows of horny scutes thathad their origin from these dermal bones. In one turtle, however, Toxochelys,of the Upper Cretaceous, there are yet remains of the dermal armor in the shape ofa row of bones along the dorsal median keel. For a description and illustrations ofthese the reader must consult later pages of this work.

In a paper written in 1898 (Amer. Naturalist, xxxii, p. 929) the writer denomi-nated such bones as the nuchal, the peripherals, and suprapygals as "fascia bones."This was done simply to distinguish them from more superficial bones, calleddermal. In a recent paper Mr. H. H. Newman (Biol. Bull, x, p. 74) has referredto this paper and has insisted that the nuchal is a dermal bone. Such it doubtless

*For Dr. George Baur's latest views regarding the primitive condition of the carapace and plastronof the turtles see Anatomischer Anzeiger, xn, 1906, p. 567.

24 FOSSIL TURTLES OF NORTH AMERICA.

originally was, but it is now overlain by the dermal mosaic. Mr. Newman holds alsothat the peripherals and the suprapygals are dermal bones, not what the writer called

fascia bones or subdermal bones. If the reader will consult the succeeding pages that

deal with Toxochelys he will see that one of the dermal nodules overhes the second

suprapygal. 1 he latter is therefore not equivalent to the row of nodules in front

of It and to those on the tail of the snapping-turtle, but is a bone of a deeper layer.

We may, therefore, for the present regard the characters of the most primitiveturtles as having been the following: There was a superficial armor composed of

dermal bones arranged in at least 7 zones above and at least 5 zones below, with

many bones in each zone, and each bone covered by a horny scute. Beneath thiswas a subdermal armor. On the upper side of the body, this consisted of the nuchal,a row of neurals, 8 pairs of costal plates, suprapygals, and peripherals; below, of

the entoplastron and at least 5 pairs of subdermal bones. There were 18 presacral

vertebrae, of which 10 belonged to the trunk and were without transverse processes

and more or less immovably joined to each other and to the ribs and neural plates.

The neck was short and consisted of 8 biconcave vertebrae, which possest transverseprocesses and low neural spines. There may possibly have been present somecervical ribs. There were 2 sacrals. The tail consisted of biconcave vertebrae,each provided with free ribs, with chevron bones below, and perhaps with one or

more rows of bony nodules above.The skull had a complete temporal roof, but the temporal fossa was probably

not widely open behind. There were nasals, lacrimals, and prefrontals, all

distinct from one another. There was no parietal foramen. A vomer was presentbut no prevomers. The choanae opened far forward in the roof of the mouth. Thequadrate was fixt, somewhat excavated to form a tympanic cavity and notchtbehind for the stapes. Probably the palate was closed by the union of the pterygoidswith each other and with the basisphenoid. The parietals may or may not havesent down each a plate to the pterygoid, in front of the exit of the trigeminal nerve.The paroccipitals were free from the exoccipitals. The jaws were without teeth andwere covered by a sheath of horn. Each ramus of the lower jaw was doubtless com-posed of 7 elements, and the dentaries were probably not co-ossified at symphysis. Theouter surface of the bones of the skull was covered by horny scutes. The shoulder-girdle consisted of a pairof coracoids, a pair of scapulae, and a pair of procoracoids;the latter probably co-ossifying somewhat late in Hfe with the scapula. Limbs, foreand hinder, probably not greatly different from those of living Chelydridae, but morecrudely modeled. The phalangeal formula was 2, 3, 3, 3, 3 in all the feet.

If the views exprest above regarding the original composition of the armor ofturtles is correct, the possession of a carapace consisting exclusively of a mosaicof dermal plates, as in Dermochelys, is a secondary character; as is likewise thepossession of a shell constituted only of the deeper elements, as in the great majority

of turtles. Other secondary characters are the absence of entoplastron {Dermo-chelys, Kinosternon), the absence of peripherals {Trionychoidea), and the absence ofmesoplastra, as in most Hving turtles. Such too is the lack of nasals and lacri-mals in the great majority of turtles. The adaptation of the hmbs for habitualswimming is, of course, a secondary modification of these organs.

In this category, too, belong those modifications of the cervical vertebrae byvirtue of which the neck has become so greatly elongated in most forms and mademost freely flexible in a horizontal plane in the Pleurodira and in a perpendicularplane in all the others. Most extraordinary of all the secondary characters is thatcomplex of modifications which permits the head and neck to be retracted betweenthe scapulae, the loop in the neck sometimes reaching backward quite to the pelvis.

CLASSIFICATION. 25

THE CLASSIFICATION OF THE TURTLES.

It is not the purpose of the writer to give here an account of the various schemesof classification that have been proposed by writers on the turtles. For such anaccount the reader may consult the third part of the sixth volume of Bronn'sKlassen und Ordnungen des Thierreichs, beginning with page 347. The followingis the arrangement ot suborders, superfamihes, and families accepted by the writer.

All of these families, except those in itahcs, have fossil representatives.

Order Testudines.Suborder I. Athecae.

Family. Dermochelyidae.

Suborder II. Thecophora.SuPERFAMiLY I. Amphichelydia.

Families. Pleurosternidae, Baenidse, Plesiochelyidae ?

SupERFAMiLY 2. Pleurodira.Families. Bothremyidae, Pelomedusidae, Chelyidae, Miolanidse.

Superfamily 3. Cryptodira.Families. Thalassemydidae, Toxochelyidas, Desmatochelyidas, Protostegidae, Cheloniids,

Tretosternidae, Chelydridae, DermatemydidaE, Platysternida, Kinosternidie

Carettochelydae, Emydidae, Testudinidae.

Superfamily 4. Trionychoidea.Families. Plastomenidae, Trionychidae.

The arrangement and names of the suborders and superfamihes above givenare the same as those used by Mr. George A. Boulenger in his Catalogue of theChelonia of the British Museum, 1889. Mr. Richard Lydekker employs practicallythe same groups in his Catalogue of the Fossil Reptilia of the British Museum,part III, 1889; but to some of his groups are given different names. Dr. LouisDoUo, of Brussels, also divides the turtles into the two suborders Athecae andThecophora. Altho this eminent writer beheves that Dermochelys, the only Hvingrepresentative of the Athecae, was derived from the Cheloniidae, he separates it asthe representative of a distinct suborder on account of its extreme modifications ofstructure. It was Cope who first proposed to make this turtle the type of a distinctsuborder.

In his Bibliography and Catalogue of the Fossil Vertebrata of North America,1902, the present writer assigned to the Trionychoidea, under the name Trionychia,the rank of a suborder. A further consideration of the subject has convinct himthat these turtles should rank lower than a suborder; not higher than a super-

family. Indeed, they appear to have brancht ofi^ from the earliest Cryptodira;

but their hneage is so ancient, and they have undergone so many modifications ofstructure, that they are of equal rank with the Cryptodira. The skull is more likethat of the Cryptodires than that of the Pleurodires, but has developt pecuharities

of its own. Like the Cryptodires, the temporal roof has never been eaten away frombelow, and always a zygomatic arch remains. The neck is wholly cryptodiran inits modifications and is retracted within the shell in the same way. This is a featureunique among animals, and it seems improbable that it could be hit upon independ-ently by two distinct groups of turtles. The pelvis in its parts and its relationshipsto the shell is entirely cryptodiran.

Ernst Haeckel, in his Systematische Phylogenie der Wirbelthiere, 1895, page 326,has taken the position that the Trionychoidea had probably arisen already in theTriassic and that they are to be lookt upon as the group from which all